RefNet
Auriel Fournier, Forest Stevens, Matt Boone, Emilio M. Bruna
2018-09-01
1. Introduction
Bibliometric databases such as the Web of Science and SCOPUS are rich sources of data for studying factors shaping the emergence, disemination, and impact of scientific research (Sugimoto and Lariviare 2018). In recent years the type and scope of questions that researchers can address with data gathered from these databases has expanded tremendously (Forutnato et al. 2018), in part because of their expanding coverage, greater accessibility, and advances in computational power that make it possible to analyze datasets comprising millions of bibliographic records (e.g., Lariviare et al. 2013, Smith et al. 2014). Nevertheless, conducting these analyses continues to be hampered by two major and persistent challenges:
The first of these challenges is Author Name Disambiguation (AND). Correctly identifying the authors of a research product is fundamental to bibliometric research, as is the ability to correctly attribute to an author all of their scholarly output. However, this seemingly simple task can often be extremely complicated, even when using the nominally high-quality data extracted from bibliometric databases (reviewed in Smalheiser and Torvik 2009). The most obvious case is when different authors have identical names, which can be quite common in some countries (Strotmann et al. 2009). However, confusion might also arise as a result of journal conventions or individual preferences for abbreviating names. For instance, one could erroneously conclude that “J.C. Smith”, “Jennifer C. Smith”, and “J. Smith” are different authors, while papers by “E. Gonzalez” could have been written be written by a single author or by two different ones whose first names start with the same letter (e.g., “Enrique”, “Eduardo”). Failure to disambiguate author names can seriously undermine the conclusions of analyses, but verifying author identity manually quickly becomes impractical as the number of authors in a dataset increases.
The second challenge is correctly parsing author addresses. The structure of author affiliations is complex and idiosyncratic, and journals differ in the information they require authors to provide as well as the way in which they present it. For instance, researchers at academic institutions might include the one or more Centers, Institutes, Colleges, Departments, or Programs in their address. The affiliations might be written in different ways in different journals (e.g., “Dept. of Biology”, “Department of Biology”, “Departamento de Biologia”); the same is true of the institution (“UC Davis”, “University of California-Davis”,“University of California”) or country in which it is based (“USA”, “United States”, “United States of America”). Researchers working for the same institution could be based at units in geographically disparate locations. Finally, these complex affiliations are all recorded in a single field of a reference’s bibliographic record. In concert, these factors can make it challenging to carry out analyses for which author affiliation or location is of particular interest.
Package refnet was written to help users of the R statistical computing environment (R Core Team 2017) address these challenges. It imports and organizes the output from Web of Science and SCOPUS searches, disambiguate author names and flags any needing additional scrutiny, parses author addresses, and georeferences author institutions. It also generates simple summary bibliometric statistics and maps author locations and coauthorships. Finally, the data can be exported in tidy formats for more in-depth analyses with user-written code or imported into packages such as ‘revtools’ (Westgate 2018a,b) or ‘bibliometrix’ (Aria & Cuccurullo 2017).
2. Using RefNet
Once bibliographic records have been downloaded (for guidance see Appendix 1), refnet’s tools are applied in four steps:
- (1) importing and tidying reference records,
- (2) parsing of author addresses and author name disambiguation,
- (3) georeferencing of author institutions, and
- (4) data analysis and visualization (Fig 1).
Each of these steps is described below; Appendix 2 guides users through the entire process with a sample dataset.
Figure 1: refnet allows users to import, organize, and visualize bibliographic data in 4 simple steps.
2.1. Importing Search Results
The refnet package can import a single Web of Science data file combine and import multiple files in the same directory. The acceptable file formats are ‘.txt’ and ‘.ciw’. For this step we will use the references_read() function, which has three arguments:
- data: The location of the directory in which the Web of Science file(s) are located. If left blank it assumes the files are in the working directory. If in a different directory, the absolute file name or relative file paths can be used.
- dir: TRUE when loading multiple files; FALSE when loading a single file. When multiple files are processed
RefNetidentifies and removes any duplicate reference records. - filename_root: The location in which the output file is to be saved and the prefix used to name it. If you do not want to write a file leave this field blank.
The output of references_read() is a an object in the R Workspace, which is also written out into a .csv file, in which each line is a WOS record and columns are the different parts of the bibliographic record (author name, journal name, article title, etc). This file is used by refnet for Step 2.
Example
- To import and process a single file located in a folder named “data” and save the output as a file named “Brazil” it in a folder named “output”:
references_read(data = './data/Brazil.txt', dir = F, filename_root = './output/Brazil')
- To import and process multiple files located in a folder named “data” and save the output as “WOS” in a folder named “output”:
references_read(data = './data', dir = T, filename_root = './output/WOS')
- To save to a location with a prefix, write the folder path followed by a ‘/’ and then the prefix. This will save the file under the root directory as “/newpath/newprefix_references.csv”:
"./newpath/newprefix".
2.2. Author address parsing and name disambiguation
The next step is to identify all unique authors in the dataset and parse the affiliation for every article they wrote. This requires identifying any authors whose name appears to be represented in different ways on different publications. Name disambiguation is a complex statistical and computational problem for which researchers have used data ranging from author affiliation to patterns of coauthorship and citation (reviewed in Smalheiser & Torvik 2009). The authors_clean() disambiguation algorithm first assigns each author of each article a unique ID number, then assigns putative name variants representing the same author a group ID number (see Appendix 3).
The function authors_clean() has two arguments:
- references: The object created by references_read(), from which author names will be extracted. Any previously generated output from references_read() that has been saved to an object can be used. - filename_root: The location in which the output file is to be saved and the prefix used to name it. If you do not want to write a file leave this field blank. The syntax is the same as for references_read().
The output of authors_clean() is two objects:
- “master”, which is the complete list of disambiguated author names, and
- “authors”, the subset of authors with putative name variants.
Users can either accept the results of the disambiguation or review the groups of names, make any necessary corrections, and merge the corrections into the master list.
Example
eb_authors <- authors_clean(eb_references, filename_root="./output/eb")
2.2.1. Accepting the results of disambiguation
If the users wish to accept the result of the disambiguation algorithm without reviewing the names results can be accepted with authors_refine(), the output of which is an object.
2.2.2. Reviewing the results of disambiguation
Users that prefer to manually review the results of the disambiguation can do so with the “authors” object and .csv file, which contain the following information:
authorID: AuthorID is a unique identifier for each name in the database (i.e., every author of every paper; the initial assumption of RefNet’s disambiguation algorithm is that all authors of all articles are different individuals).
AU:
AF:
groupID: This indicates which names (i.e., AuthorID numbers) have been grouped together under a single groupID number because they are believed to be the same person. Because disambiguation is performed iteratively. The lowest authorID number in a group will always be used as the groupID.
match_name: The name under which the algorithm groups all of an author’s putative name variants.
similarity: The Jaro-Winkler similarity score when comparing —- and — (range = 1-100, with a score of 100 indicating an exact match). We do not perform this analysis on name that are identical or matches without similar last name, first name, middle combinations. [this needs a better explanation]
author_order
address: the author’s address in the WOS record for an article.
university - the author’s institution, if present from the address
department - the author’s department, if present, from the address
short_address - the author’s street address
postal_code - the author’s postal code
country - the author’s country
RP_address - the reprint address, if present
RI: the author’s Thomson-Reuters Researcher ID number in the WOS record for an article(if they have one).
OI: the author’s ORCID ID number in the WOS record for an article (if they have one).
EM:the author’s email address in the WOS record for an article (if it lists one).
UT: Accession Number
refID
PT: Publication Type (J=Journal; B=Book; S=Series; P=Patent)
PY: Publication Year
PU
Users can assess if the validity of grouping different author name variants under a single groupID using the similarity score, author address, ORCID AND Researcher ID numbers, and author email address. They can also use the original record for the authors and articles to search for them online.
2.2.3. Correcting the results of disambiguation
To correct cases where authors that are NOT the same person were erroneously assigned the same groupID, simply replace the number in the groupID column with the value from the authorID column (be sure to use the authorID value from the same row).
To correct cases where names weren’t grouped together but should have been (i.e., they represent variants of the same author’s name), replace the value in the GroupID column of each name to be grouped together with a single AuthorID number. We recommend using the lowest authorID number of the name variants being pooled.
Once the revision is complete, save the file as a .csv file [EB: with the same name?] in the same location as the original file.
2.2.4. Merging corrections
Corrections made to the “Author” file are merged into the “Master List” file using the authors_refine() function. It has four arguments:
- authors: The corrected version of the “authors” object. Must be in an object.
- master: The unedited master file created with
authors_clean(). Must be in an object
- sim_score: The threshold for the similarity score below which authors are assigned different groupIDs. By default this is turned off.
- filename_root: The prefix for the resulting output and locations to which it will be saved, using the same syntax as for previous functions.
The output of authors_refine() is a file. This file can be used for many analyses, including.
Example
eb_authors <- authors_refine(eb_authors, eb_authors_master, filename_root="./output/eb")
4. Bibliometric analyses and Data Visualization
RefNet includes some basic summary analyses and data visualization tools….
4.1 Bibliometric analyses EB: I HAVE CODE TO DO ALL OF THIS!
- OVERALL
- total number of articles
- total number of authors
- Average number of authors per article (+/-SD or SE)
- Histogram of authors per article
- total number of countries
- BY YEAR
- total number of articles
- total number of authors
- cumulative number of authors
- average number of authors per article (+/-SD or SE)
- number of countries per year
- cumulative number of countries
4.2 Visualizing Productivity and Collaboration
There five kinds of plots you can make. They are in two categories World plots and Net Plots. World plots simply plot author locations on a world map. Net plots use network analysis to plot connections of co-authorship on papers. We’ll go from simple to complex. It’s important to note that the rendering of these plots may take time, particularly if the amount of individual authors is >1000, depending on your computers processing power. So give it time if you are using an older computer or attempting to map a large number of points
4.2.1. World Plots (by country)
To make a colored plot of all authors country of residency we’ll use the plot_addresses_country(). The function has one argument:
- data: This is the output from addresses_lat_long. Must be an object.
4.2.2. World Plots (points)
To individual plot points of each authors location we’ll use the plot_addresses_points(). The function has one argument:
- data: This is the output from
authors_georef(). Must be an object.
4.2.3. Net Plots (base)
To plot a network diagram of co-author countries of origins and how they’re connected we’ll use the net_plot_coauthor() function. The function has one argument:
- data: This is the output from
authors_georef(). Must be an object.
This function has two outputs, in the form of a list. $plot is the plot, $data is the dataframe that was used to the build the plot. listname$plot is not very customizable, which is why $data is provided, so the user can build their own plot to their specific needs.
4.2.4. Net Plots (Country)
To plot a network diagram of co-author countries of origins overlayed on a world map we’ll use net_plot_coauthor_country() function. The function has one argument:
- data: This is the output from
authors_georef(). Must be an object.
The net plot analysis returns an object that allows the plotting of these complex connections. $plot will plot the output to your plotting console. $data is the dataframe that was used to the build the plot. $plot is not very customizable, which is why $data is provided, so the user can build their own plot to their specific needs.
4.2.5. Net Plots (Addresses)
To plot a network diagram of co-author using their individual addresses overlayed on a world map we’ll use net_plot_coauthor_addresses() function. The function has one argument:
- data: This is the output from
authors_georef(). Must be an object.
The net plot analysis returns an object that allows the plotting of these complex connections. $plot will plot the output to your plotting console. $data is the dataframe that was used to the build the plot. $plot is not very customizable, which is why $data is provided, so the user can build their own plot to their specific needs.
Summary
Acknowledgments
Support for the development of RefNet was provided by grants from the University of Florida Center for Latin American Studies and the University of Florida Informatics Institute.
References
Aria, M. & Cuccurullo, C. (2017) bibliometrix: An R-tool for comprehensive science mapping analysis. Journal of Informetrics. 11(4): 959-975.
Fortunato, S., C. T. Bergstrom, K. Barner, J. A. Evans, D. Helbing, S. Milojevic, A. M. Petersen, F. Radicchi, R. Sinatra, B. Uzzi, A. Vespignani, L. Waltman, D. Wang, & A.-L. Barabasi (2018). Science of science. Science, 359:eaao0185.
Lariviare, V., Ni, C., Gingras, Y., Cronin, B., & Sugimoto, C. R. (2013). Bibliometrics: Global gender disparities in science. Nature News, 504(7479): 211-213
R Core Team (2017). R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria. URL https://www.R-project.org/.
Smalheiser, N. R., & Torvik, V. I. (2009). Author name disambiguation. Annual Review of Information Science and Technology 43(1): 1-43.
Smith, M. J., Weinberger, C., Bruna, E. M., & Allesina, S. (2014). The scientific impact of nations: Journal placement and citation performance. PLoS One 9(10): e109195.
Strotmann, A. and Zhao, D., (2012). Author name disambiguation: What difference does it make in author based citation analysis?. Journal of the Association for Information Science and Technology. 63(9): 1820-1833.
Sugimoto CR, Lariviare V. (2018). Measuring Research: What Everyone Needs to Know®. Oxford University Press, Oxford, UK. 149 pp.
Sci2 Team. (2009). Science of Science (Sci2) Tool. Indiana University and SciTech Strategies, https://sci2.cns.iu.edu.
Westgate, M. J. (2018a). revtools: Tools to Support Evidence Synthesis. R package version 0.2.2. https://CRAN.R-project.org/package=revtools
Westgate, M. J. (2018b). revtools: bibliographic data visualization for evidence synthesis in R. bioRxiv:262881. doi: 10.1101/262881
Appendix 1: Giode to downloading reference records from the Web of Science©.
